IKZF3 Deficiency Potentiates Chimeric Antigen Receptor T Cells Targeting Solid Tumors

bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 IKZF3 deficiency potentiates chimeric antigen receptor T cells targeting solid tumors

2 Yan Zou1, Tian Chi2,3*

4 1 Shanghai Institute for Advanced Immunochemical Studies, ShanghaiTech University, Shanghai

5 201210, China.

7 2 School of Life Science and Technology, ShanghaiTech University, 393 Middle Huaxia Road, Pudong

8 New Area, Shanghai 201210, China

9 3 Department of Immunobiology, Yale University Medical School, New Haven, CT, USA

11 *Correspondence: Tian Chi, School of Life Science and Technology, ShanghaiTech University, 393

12 Middle Huaxia Road, Pudong, 201210 Shanghai, China. E-mail: chitian@ shanghaitech.edu.cn

14 Short title: IKZF3 deficiency potentiates CAR T cells

1 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

16 Abstract

17 Chimeric antigen receptor (CAR) T cell therapy has been successful in treating hematological

18 malignancy, but solid tumors remain refractory. Here, we demonstrated that knocking out transcription

19 factor IKZF3 in HER2-specific CAR T cells targeting breast cancer cells did not affect proliferation or

20 differentiation of the CAR T cells in the absence of tumors, but markedly enhanced killing of the

21 cancer cells in vitro and in a xenograft model. Furthermore, IKZF3 KO had similar effects on the

22 CD133-specific CAR T cells targeting glioblastoma cells. AlphaLISA and RNA-seq analyses indicate

23 that IKZF3 KO increased the expression of genes involved in cytokine signaling, chemotaxis and

24 cytotoxicity. Our results suggest a general strategy for enhancing CAR T efficacy on solid tumors.

2 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

26 Introduction

27 Chimeric Antigen Receptor (CAR) T cells are human primary T cells genetically modified to express a

28 CAR, the latter comprising the extracellular single-chain variable fragment (scFv) targeting surface-

29 exposed tumor-associated antigens (TAA) in non-MHC restricted manner, a transmembrane domain,

30 and one or more intracellular signaling domains that activate T cells upon antigen engagement1. CAR T

31 cells are effective in eradicating some hematological malignancies2-5 . However, despite much effort,

32 solid tumors have proved largely unresponsive due to inefficient infiltration of CAR T cells, presence

33 of immunosuppressive tumor microenvironment and exhaustion of T cells1, 6, 7. Various strategies have

34 been attempted to potentiate CAR-T cells targeting solid tumors, including the expression of

35 chemokine receptors8-10, disruption of the inhibitory adenosine and TGF-β signaling via deletion of

36 adenosine and TGF-β receptor A2AR and TGFBR2, respectively11, 12, administration of IL-7, IL-15 or

37 IL-2113-15, use of immune checkpoint inhibitors16-18, or deletion of PD-1 gene in CAR T cells19-21, but

38 their benefits are limited. For example, the use of immune checkpoint inhibitors in conjunction with

39 CAR T therapy may increase the risk of autoimmunity, while PD-1 knockout may lead to the failure of

40 CD8+ T cells, as PD-1 expression can protect CD8+ T cells from excessive proliferation and terminal

41 differentiation22. It is thus highly desirable to devise additional methods to harness the power of CAR

42 T cells for solid tumor treatment.

44 Lenalidomide (LENA) is an immunomodulatory drug with pleotropic effects on diverse immune cells

45 including natural killer cells, monocytes, B cells and T cells23-25. In T cells, LENA can increase IL-2

46 induction upon TCR stimulation26, shift T helper responses from Th2 to Th127, and inhibit Treg

47 function28. LENA acts mainly by causing the ubiquitin-mediated degradation of the Ikaros family zinc

48 finger transcription factor IKZF1 and IKZF329, 30. In T cells, siRNA-mediated knockdown of either

49 IKZF1 or IKZF3 is sufficient to enhance the expression of IL-2 and multiple other cytokines,

50 reminiscent of the effects of LENA treatment29. We have recently demonstrated that in vitro, LENA

51 can enhance cytokine expression and killing of solid tumors by CAR T cells, suggesting that LENA

52 might be used for boosting the efficacy of CAR T cells targeting solid tumors31. However, LENA has

53 been FDA-approved only for treating several hematological cancers (namely multiple myeloma,

54 myelodysplastic syndromes and mantle cell lymphoma), but not for any solid tumors. One potential

55 strategy to overcome this barrier would be to create CAR T cells deficient in IKZF1 and/or IKZF3, 3 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

56 which may recapitulate the beneficial effects of LENA on T cells but can avoid the unintended

57 pleiotropic effects on other cell types that would result from systemic administration of LENA. In this

58 study, we have opted to target IKZF3 instead of IKZF1 in CAR T cells, because IKZF1 has the

59 characteristics of a tumor suppressor gene: IKZF1 can arrest the uncontrolled growth and proliferation

60 when introduced into an established mouse Ikaros null T-leukemia cell line32-34, and IKZF1 promotes

61 the development of myeloproliferative neoplasms35. In contrast, such effects have not been described

62 for IKZF3, suggesting it is safer to knock out IKZF3 in T cells as compared with IKZF1. Our results

63 demonstrate that IKZF3 KO markedly enhanced the ability of CAR T cells to kill solid tumor cells in

64 vitro and in a xenograft model, which was associated with increased expression of genes mediating

65 cytokine signaling, cell trafficking and cytotoxicity. These data suggest a novel strategy for boosting

66 the efficacy of CAR T therapy for solid tumors.

68 Results and Discussion

69 Generation of IKZF3-deficient HER2-specific CAR T cells

70 The CAR T cells we sought to optimize expressed HER2-targeting CAR comprising a Myc-tagged

71 scFv derived from the Herceptin monoclonal antibody, the CD8 transmembrane domain upstream of

72 the tandem signaling domains from CD28, 4-1BB and CD3-ζ36, which is a commonly used form of

73 CAR37. To generate IKZF3-deficient HER2-CAR T cells (termed HER2K3-CAR T), we nucleofected

74 PBMCs from two healthy donors with expression vectors for CAR, piggyBac transposase, Cas9 and

75 the dual gRNAs targeting IKZF3 exon3; the dual gRNA format was used due to higher editing

76 efficiency38. Control CAR T cells (termed HER2-CAR T) were generated in parallel by omitting Cas9

77 and gRNA expression vectors. Nucleofected cells were then cultured with irradiated allogenic PBMCs,

78 anti-CD3 antibody, IL-2 and puromycin to expand and select CAR-T cells. The cells at the end of the

79 expansion were then characterized.

81 We found that over 89% and 79% of the Cas9/gRNA transfected CAR T cells from Donor 0529 and

82 Donor 0710, respectively, harbored out-of-frame mutations at IKZF3 exon 3 (Figure 1A) concomitant

83 with marked reduction in IKZF3 protein expression (Figure 1B), thus demonstrating highly efficient

84 editing. Proliferation of these cells was somewhat impaired by IKZF3 disruption (Figure 1C) but

85 importantly, CAR expression was unaffected (Figure 1D-E), and neither was the phenotype related to 4 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

86 cell differentiation altered as defined by CD62L and CD45RO expression (Figure 1F-G; note that

87 donor-specific differences in the relative abundance of central vs effector memory cells). Thus, IKZF3

88 was dispensable for CAR expression or cell differentiation during the expansion and selection of the

89 CAR T cells. However, IKZF3 deficiency induced CD69 expression in CAR T cells cocultured with

90 allogeneic PBMCs (Figure 1H-I), indicating IKZF3 involved in TCR stimulation signals downstream

91 of allogeneic response.

93 IKZF3 deficiency potentiates killing of breast cancer by HER2-specific CAR T cells

94 We next investigated the cytolytic efficacy of the HER2K3-CAR T cells using three assays.

96 First, the CAR T cells were incubated with MDA-MB-453 cells (originating from a human mammary

97 gland cancer) expressing firefly luciferase (termed “453-ffluc” hereafter) for 48 h, and the loss of

98 luciferase activity within viable cells was taken as a measure of cell lysis. HER2K3-CAR T cells

99 showed enhanced killing compared with HER2-CAR T cells (Figure 2A), which was correlated with

100 augmented production of IFN-γ, IL-2, TNF-α and GM-CSF (Figure 2B). We also determined the

101 proliferation of the CAR T cells by culturing CFSE-labeled CAR T cells with irradiated MDA-MB-453

102 cells. After seven days of culture, the HER2K3-CAR T population harbored more divided cells than

103 HER2-CAR T (78.5% vs. 62.7%) and had undergone more rounds of division (1.4 vs 1; Figure 2C),

104 which correlated with higher frequency of CD69 expression (Figure 1H-I), indicating that IKZF3

105 deficiency facilitated tumor-stimulated CAR T cell activation and proliferation.

106

107 Second, impedance-based killing assays were carried out to assess the cytolytic activity over a longer

108 period of time. Specifically, MDA-MB-453 cells were attached to the electrode plate, and CAR T cells

109 then added to lyse and detach the tumor cells, which was read out as changes in the electrical

110 impedance. We measured the impedance every 15 min over a 150 h period, and also determined

111 cytokine concentration in the media at the end of the experiment. HER2K3-CAR T cells were more

112 effective at tumor killing at various effector to target ratios over the entire course of the experiment

113 (Figure 2D), correlated with higher IFN-γ and TNF-α concentration at the end of the experiment,

114 although by this time, GM-CSF concentrations had become comparable in both groups of T cells while

5 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

115 IL-2 had become undetectable (Figure 2E and data not shown). The same trend was observed for the

116 CAR T cells derived from the second donor (Figure S1).

117

118 Finally, we assayed tumor killing in vivo. Immunodeficient NPSG mice were subcutaneously

119 inoculated with 1.5 x 106 453-ffluc cells. Six days later, 5 x 106 CAR T cells were injected i.v. and

120 tumor growth monitored using bioluminescence imaging at various times thereafter. HER2K3-CAR T

121 cells proved more efficient at tumor suppression, with bioluminescence eliminated in the 3 of 5 mice

122 (as opposed to 1 out of 5 mice for HER2-CAR T; Figure 3A, 3B), leading to enhanced mice survival

123 (Figure 3C).

124

125 We conclude that HER2K3-CAR T cells were more potent than HER2-CAR T cells at killing MDA-

126 MB-453 tumor cells, which might result from enhanced activation, proliferation and cytokine

127 production following tumor stimulation.

128

129 IKZF3 deficiency upregulates genes important for T cell function

130 To further understand how IKZF3 deficiency augmented CAR T cell function, we used RNA-seq to

131 compare gene expression patterns in HER2- CAR T vs. HER2K3-CAR T cells, finding IKZF3 KO

132 upregulated 106 genes and downregulated 92 genes (excluding genes encoding TCR variable regions;

133 Fig. 4A), which contained 97 and 86 protein-coding genes, respectively (Supplementary Table).

134 Multiple biologically interesting pathways were detected among the 97 upregulated genes, which

135 jointly suggested that IKZF3 KO enhanced cytokine signaling, chemotaxis, adhesion and immune

136 responses, consistent with the potentiation in the killing ability (Fig. 4B). IFNG and CSF2/GM-CSF

137 were repetitively enriched in multiple pathways (Categories 1, 5-8, red), suggesting their functional

138 importance. The extents of the upregulation of IFNG and CSF2/GM-CSF in the KO cells, as revealed

139 by manual examination of the RNA-seq data, were moderate (2-3 folds, Fig. 4C, top left), echoing the

140 results of the AlphaLISA assay (Fig. 2B). In addition, the transcription factor Eomes was also

141 upregulated (Fig. 4B, Category 6), as was TBX21 (Fig. 4C, top right). Both factors are important

142 regulators of CD8 T cell function39, with Eomes upregulated in effector CD8 cells and contributing to

143 IFN-γ production and cytotoxicity of CD8 T cells40, 41, while TBX21 is upregulated and plays

144 important roles in memory CD8 T cells. It is also noteworthy that genes capable of apoptosis regulation 6 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

145 were also upregulated (Categories 4, 7), which included GZMB important for T cell cytotoxicity (Fig.

146 4B, pink gene in Category 7). Remarkably, manual examination of the RNA-seq data indicates that the

147 remaining members of GZM family, namely GZMA, GZMH, GZMK and GZMM, all tended to be

148 overexpressed in the KO cells except GZMM (Fig. 4C, bottom). Furthermore, Perforin and the

149 degranulation marker CD107a were also upregulated (Fig. 4C, bottom). Thus, enhanced killing might

150 also result from overexpression and degranulation of granzymes and perforin in CD8 T cells.

151 Paradoxically, the apoptosis-related genes overexpressed in the KO cells included the proapoptotic

152 genes BCL2L11 and CDKN2A, with unclear biological relevance (Categories 4 and 7, blue).

153

154 In contrast to the upregulated genes, for the 86 downregulated genes, all pathways detected had

155 relatively large FDR (>0.049), and only 4 pathways had FDR<0.1 (Fig. 4B, bottom). The relevance of

156 these 4 enriched pathways to CAR T function was unclear. Of note, S1PR1, a pro-survival gene, was

157 present in 3 out of the 4 pathways (Fig. 4B, bottom, blue). S1PR1 downregulation reinforced the notion

158 that the KO cells might be apoptotic, as first suggested by BCL2L11 and CDKN2A upregulation.

159

160 We conclude that IKZF3 deficiency in HER2-CAR T cells enhanced the expression of genes

161 promoting cytokine signaling, chemotaxis and cytotoxicity, which might underlie the potentiation of

162 CAR T therapeutic effects. However, IKZF3 deficiency might also cause a “side-effect”, namely an

163 increase in CAR T apoptosis. Countering this putative effect could perhaps further improve the

164 function of the KO CAR T cells.

165

166 IKZF3 deficiency potentiates CD133-specific CAR T cells

167 Finally, we determined whether IKZF3 KO could potentiate CAR T cells targeting a different antigen

168 such as CD133, a biomarker for tumor-initiating cells in multiple human cancers, especially

169 glioblastoma42, where a small population of CD133 positive glioblastoma tumor-initiating cells manage

170 to survive radiotherapy and chemotherapy, conferring resistance to the therapies43, 44.

171

172 CD133-specific CAR (133-CAR) T cells and their IKZF3-deficient (133K3-CAR) counterpart were

173 generated from two donors as in the case of the HER2-specific CAR T cells. In the 133K3-CAR T cells

174 from both donors, the IKZF3 gene was efficiently mutated (Figure 5A) and the protein eliminated 7 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

175 (Figure 5B). IKZF3 -deficiency did not affect cell expansion (Figure 5C) or CAR expression (Figure

176 5D, 5E), but for unknown reasons, enriched or partially depleted the CD62L+ CD45RO+ central

177 memory T cell subset depending on the T cell donors (Figure 5F-G). CD69 expression was increased in

178 IKZF3 KO (Figure 5H-I) as indicated in HER2-CAR T cells (Figure 1H-I). Importantly, for both

179 donors, IKZF3 -deficiency potentiated killing of the glioblastoma U251 cells overexpressing CD133

180 (U251-OE-ffluc) in short-term cultures (Figure 6A; Figure S2A), concomitant with increased cytokine

181 production (Figure 6B; Figure S2B), cell proliferation (Figure 6C) and CD69 induction (Figure 5H-I;

182 Figure S2C). In long-term (160 h) cultures, IKZF3 -deficiency similarly potentiated killing (Figure 6D;

183 Figure S2D), which was associated with increased production of IFN-γ and TNF-α (but not IL-2 or

184 GM-CSF; Figure 6E). Thus, the effects of IKZF3 elimination on CD133-specific CAR T cells were

185 similar to that on HER2-CAR T cells.

186

187 Conclusion

188 IKZF3 KO in CAR T cells can markedly enhance their therapeutic effects in treating different types of

189 solid tumors, suggesting that IKZF3 KO is an attractive alternative to the use of LENA in enhancing

190 CAR T functions.

191

192

193

194

195

8 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

196 Materials and Methods

197 Plasmids

198 PiggyBac transposon vector expressing CD133-specific CAR and the plasmid expressing piggyBac

199 transposase (PBase) were prepared according to Zhu et al45. HER2-specific CAR expression construct

200 was generated by replacing CD133-specific scFv with the HER2-specific scFv derived from the mAb

201 Herceptin. The Cas9 expression plasmid pST1374-Cas9-GFP was created according to Su et al38. The

202 sgRNA pair targeting the IKZF3 gene locus was designed using E-CRISP website (http://www.e-

203 crisp.org/E-CRISP/) and expressed from pGL3-U6 sgRNA-PGK-Puro vector (Addgene 51133).

204

205 Tumor cell lines

206 Glioma U251 and breast cancer MDA-MB-453 lines were purchased from Cell Bank of the Chinese

207 Academy of Sciences (Shanghai, China) and cultured in DMEM containing 10% FBS and 1%

208 penicillin/ streptomycin. The cells were routinely checked for mycoplasma infection. To generate the

209 U251-CD133OE line, the piggyBac transposon vector (2.5 μg) expressing the human CD133 antigen45

210 was co-delivered with a PBase expression vector (2.5 μg) into 2 x 106 U251 cells using Nucleofector

211 2b (Lonza, Köln, Germany), followed by puromycin ( 500 ng/ml) selection of the stably transfected

212 cells for 14 days. To generate the U251-CD133OE-ffluc and 453-ffluc lines, 2 x 106 U251-CD133OE

213 and MDA-MB-453 cells were nucleofected with a piggyBac transposon vector (2.5 μg) expressing

214 firefly luciferase with the PBase vector (2.5 μg) and the stably transfected cells selected on G418 ( 500

215 μg/ml) for 14 days.

216

217 CAR-T cell generation and expansion

218 PBMCs from healthy donors were purchased from HemaCare (PB009C-3). To generate IKZF3-

219 deficient CAR-T cells, 15-20 million PBMCs in RPMI 1640 (10% FBS, 1% penicillin/streptomycin)

220 were nucleofected using Nucleofector 2b (Lonza, Köln, Germany) with a mixture of 4 plasmids (5 μg

221 each) expressing CAR, PBase, gRNAs and Cas9, respectively. IKZF3-sufficient control CAR-T cells

222 were generated in the same way except that the gRNAs and Cas9 expression vectors were omitted

223 during nucleofection.

224 24 h following nucleofection, a mixture of allogenic PBMCs derived from 5 donors , irradiated at a

225 dose of 40 Gy with an X‐ray irradiator (Rad Source Technologies), were added to the nucleofected 9 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

226 cells at the ratio of 15:1 in the presence of anti-CD3 antibody (50 ng/ml , 130-093-387, Miltenyi

227 Biotec) and IL-2 (300 IU/ml, Novoprotein) as described 45. 48 h later, the culture media was replaced

228 with AIM V containing 10% fetal bovine serum (Gibco) and 300 IU/ml IL-2 (Novoprotein). Puromycin

229 (0.5 μg/ml) was added and CAR-expressing T cells selected for at least 7 days. CAR expression and

230 activation/differentiation of the expanding T cells were monitored by flow cytometry (CytoFLEX

231 cytometer, Beckman Coulter) twice a week (see further). 14 days after nucleofection, the expanded

232 CAR T cells were re-stimulated with PBMC and anti-CD3 as in the first-round stimulation and further

233 expanded for 10 days.

234

235 Flow cytometry

236 CAR-T cells were stained with anti-Myc antibody (clone 9B11, 2276S, Cell Signaling Technology) to

237 detect Myc-tagged CAR, with anti-CD45RO (559865, BD Biosciences) and anti-CD62L (562719, BD

238 Biosciences) to evaluate differentiation into memory and effector cells, and with anti-CD69 (555533,

239 BD Biosciences) to assess activation. CD133 and HER2 expression on the tumor cell lines were

240 detected using anti-CD133 (130-098-826, Miltenyi Biotec) and anti-HER2 (BMS120Fl, ThermoFisher

241 Scientific), respectively. All samples were washed with FACS buffer (PBS with 0.5% BSA and 2 mM

242 EDTA) before addition of antibodies. After incubation in 4℃ for 30 min, cells were washed with

243 FACS buffer prior to analysis on the flow cytometer. We collected the data on 10000 events and in all

244 cases appropriate isotypes were used as negative controls. Flow cytometry was performed on a

245 CytoFLEX cytometer (Beckman Coulter), and analyzed using FlowJo software.

246

247 DNA extraction, genotyping and sequencing

248 Genomic DNA was extracted using DNeasy® Blood & Tissue Kit (QIAGEN) before PCR

249 amplification of the targeted site at the IKZF3 locus using the following primers:

250 Forward GAGAACCCTTCCTCTCCCCT; Reverse ACGTGGCTGCATTAGGAGAG

251 The resulting PCR amplicons (658 bp) were cloned into pGEMT vector (Promega) before Sanger

252 sequencing of individual inserts.

253

254 Western blot analysis of IKZF3 expression

10 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

255 2 x 106 CAR T cells collected after 24-day expansion were resuspended with 1 x Laemmli sample

256 buffer (BIORAD) containing 2.5% β-mercaptoethanol before sonication with 50% amplitude in Sonic

257 Dismembrator (FisherBrand). The lysate was heated at 95℃ for 5 min and centrifuged at 10000 rpm/s

258 for 2 min to remove the insoluble material prior to SDS-PAGE in a 4-20% gradient ExpressPlusTM

259 PAGE Gels (Genscript) at constant 120 V for 1 h. The proteins were transferred to PVDF membranes

260 with Semi-Dry Electrophoretic Transfer Cell (BIORAD) at constant 15V for 30min. The membranes

261 were blocked by 5% BSA in PBS with 0.05% Tween-20, followed by incubation with rabbit anti-

262 IKZF3 (NPB2-24495, Novus) or mouse anti-GAPDH (sc-32233, Santa Cruz) antibodies in 4℃

263 overnight. Membranes were then washed in PBST, followed by 1h incubation with HRP-conjugated

264 donkey-anti-rabbit antibody (711-035-152, Jackson) or donkey-anti-mouse antibody (711-035-150,

265 Jackson) for IKZF3 or GAPDH protein, respectively. After washing in PBST, the proteins were

266 visualized with Chemiluminescent HRP Substrate (Merk Millipore) and signals recorded using Bio-

267 Rad ChemiDoc MP chemiluminescence imaging system (BIORAD).

268

269 Luciferase-based cytolytic assay

270 Firefly luciferase-expressing tumor cells were plated in triplicates in opaque white walled plate

271 (Corning), and CAR T cells were added according to desired effector to target ratios. 48 h (for HER2-

272 specific CAR T) or 72 h (for CD133-specific CAR T cells) later, D-Luciferin potassium salt

273 (PerkinElmer) dissolved in sterile water was added at the final concentration of 150 μg/ml. The

274 Bioluminescence signal was recorded using the EnVision® Multimode Plate Reader (PerkinElmer) and

275 the percentage of tumor lysis was calculated according to the following formula: Tumor cells lysis (%)

276 = 100% × (1 – signal co-culture experiment / signal tumor alone).

277

278 Impedance-based real time cytolytic assay

279 The cytolytic capability of CAR T cells over a 180-h period was monitored using xCELLigence RTCA

280 system (ACEA Biosciences). MDA-MB-453 or U251-CD133OE were plated in E-plate (ACEA

281 Biosciences) in quadruplicate at 5000 or 10000 cells per well, respectively. After 24 h, CAR T cells

282 were added according to desired effector to target ratios. Lysed tumor cells detached from the bottom

283 of the electrode plate and changed electrical impedance, which was recorded by the system. The

11 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

284 impedance-based cell index was measured in 15 min intervals, and individual cell index was

285 normalized with the cell index prior to the addition of CAR T cells.

286

287 Cytokine secretion assay

288 Tumor cells and CAR T cells were co-cultured according to desired effector to target ratios in

289 triplicates for 24 h. The supernatant was collected and various cytokines measured using the

290 PerkinElme AlphaLISA kit (IL-2, AL221C; IFN-γ, AL217C; TNF-α, AL208C; GM-CSF, AL216C).

291

292 T cell proliferation assay

293 1-3x106 HER2-specific CAR T cells were labeled with 0.5 μM CFSE (eBioscience) and cultured with

294 equal numbers of irradiated MDA-MB-453 cells (70 Gy via X‐ray irradiator, Rad Source

295 Technologies) for 7 days in RPMI containing 10% FBS, 1% penicillin/streptomycin. 1 x 105 cells were

296 analyzed for CFSE signal by flow cytometry before and at the end of culture. The data were analyzed

297 by Flowjo to determine the Percent Divided, the value for the percentage of cells that divided at least

298 once, and the Division Index, the average number of divisions for all cells in the original population.

299 The proliferation of CD133-specific CAR T cells was analyzed in the same way except that half the

300 amounts of U251-CD133OE cells were used and the cells cultured for 4 days.

301

302 Xenograft model and tumor growth in vivo

303 Animal experiments were performed at the National Center for Protein Science in Shanghai following

304 the guidelines of the Shanghai Administrative Committee for Laboratory Animals. 1.5 x 106 453-ffluc

305 cells in 100 μl sterile PBS were mixed with equal volume of ice-cold Matrigel® (Corning). 200 μl of

306 the mixture was subcutaneously injected over the dorsal flank of the right foreleg into 6-week-old

307 NPSG females (NOD‐Prkdcscid Il2rgnull, Shanghai Jihui Laboratory Animal Care Co.,Ltd.). Six days

308 later, tumor-bearing mice were randomly grouped and intravenously injected with HER2-CAR,

309 HER2K3-CAR T cells (each at 5 x 106 cells in PBS per mouse) or PBS. Tumor growth was monitored

310 by measuring the tumor volumes and imaging. Tumor volumes, defined as (length x width2)/2, were

311 determined using a caliper. To measure the bioluminescence signal, the mice were anesthetized via

312 administration of 2% isoflurane with Matrx VIP 3000 Calibrated Vapoizer (MIDMARK) at 2 L/min

313 flow rate of total input oxygen. D-Luciferin potassium salt (PerkinElmer) was injected intraperitoneally 12 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

314 (150 mg/kg), and tumors imaged with the IVIS Lumina II device (PerkinElmer) 10 min afterwards. The

315 imaging was performed twice a week 6 days after the subcutaneous inoculation of tumor cells. The

316 mice were sacrificed when body weight loss reached 20% or the tumor volume reached 1000 mm3.

317

318 RNA-seq

319 After expansion and selection for 24 days, 1 x 106 IKZF3 knockout and parental HER2-specific CAR T

320 cells were collected. Total RNA was isolated by phenol-chloroform extraction method and its integrity

321 evaluated using the Agilent 2100 Bioanalyzer (Agilent Technologies). The libraries were constructed

322 using TruSeq Stranded mRNA LTSample Prep Kit (Illumina), followed by sequencing on the illumine

323 sequencing platform (HiSeqTM 2500, Illumina, Shanghai OE Biotech Co., Ltd.), and 150-bp paired-

324 end reads were generated. All the raw data were checked with fastQC

325 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/) for quality control. Fastq files were

326 processed with Salmon46 and the reads were mapped to human hg38 transcriptome. The data was

327 subsequently uploaded into iDEP (http://bioinformatics.sdstate.edu/idep/) to analyze differentially

328 expressed genes (DEGs). Enrichment analysis for the DEGs was performed using DAVID

329 Bioinformatics Resources 6.8 (https://david.ncifcrf.gov/). The RNA sequencing data has been

330 deposited in the NCBI database with accession number PRJNA684699.

331

13 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

332 Acknowledgements

333 The authors would like to thank the HTS platform in SIAIS and the Molecular and Cell Biology Core

334 Facility (MCBCF) at the School of Life Science and Technology, ShanghaiTech University, and the

335 Animal Facility at the National Facility for Protein Science in Shanghai (NFPS), Zhangjiang Lab,

336 Shanghai Advanced Research Institute, Chinese Academy of Science for providing technical support.

337 This work was supported by the National Key R&D Program (2019YFA0111001) of China.

338

339 Supplemental Information

340 Table 1 contains sequence of primers for PCR amplification of targeting sites.

341 An excel file listing all the differentially expressed genes (FC>=1.5, q<=0.1) is included.

342

343 Author contributions

344 Y.Z., T.C. and X.Z. designed the research. Y.Z. performed the experiments. Y.Z., T.C. and X.Z.

345 analyzed the data. Y.Z., T.C., Z.J. and Q.Y. performed the RNA sequencing analyses. L.L. and J.T.

346 coordinated the research. Y.Z. and T.C. wrote the manuscript. T.C., X.H. and X.Z. revised the data and

347 performed a final revision of the paper.

348

349 Conflicts of Interest

350 The authors declare no conflict of interest.

14 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

351 1. Newick, K, O'Brien, S, Moon, E, and Albelda, SM (2017). CAR T cell therapy for solid tumors. 352 Annu Rev Med 68: 139-152. 353 2. Kalos, M, Levine, BL, Porter, DL, Katz, S, Grupp, SA, Bagg, A, et al. (2011). T cells with chimeric 354 antigen receptors have potent antitumor effects and can establish memory in patients with 355 advanced leukemia. Sci Transl Med 3: 95ra73. 356 3. Cao, J, Wang, G, Cheng, H, Wei, C, Qi, K, Sang, W, et al. (2018). Potent anti-leukemia activities 357 of humanized CD19-targeted Chimeric antigen receptor T (CAR-T) cells in patients with 358 relapsed/refractory acute lymphoblastic leukemia. Am J Hematol 93: 851-858. 359 4. Gill, S, Maus, MV, and Porter, DL (2016). Chimeric antigen receptor T cell therapy: 25years in 360 the making. Blood Rev 30: 157-167. 361 5. Wei, G, Ding, L, Wang, J, Hu, Y, and Huang, H (2017). Advances of CD19-directed chimeric 362 antigen receptor-modified T cells in refractory/relapsed acute lymphoblastic leukemia. Exp 363 Hematol Oncol 6: 10. 364 6. D'Aloia, MM, Zizzari, IG, Sacchetti, B, Pierelli, L, and Alimandi, M (2018). CAR-T cells: the long 365 and winding road to solid tumors. Cell Death Dis 9: 282. 366 7. Martinez, M, and Moon, EK (2019). CAR T cells for solid tumors: new strategies for finding, 367 infiltrating, and surviving in the tumor microenvironment. Front Immunol 10: 128. 368 8. Craddock, JA, Lu, A, Bear, A, Pule, M, Brenner, MK, Rooney, CM, et al. (2010). Enhanced 369 tumor trafficking of GD2 chimeric antigen receptor T cells by expression of the chemokine 370 receptor CCR2b. J Immunother 33: 780-788. 371 9. Moon, EK, Carpenito, C, Sun, J, Wang, LC, Kapoor, V, Predina, J, et al. (2011). Expression of a 372 functional CCR2 receptor enhances tumor localization and tumor eradication by retargeted 373 human T cells expressing a mesothelin-specific chimeric antibody receptor. Clin Cancer Res 374 17: 4719-4730. 375 10. Di Stasi, A, De Angelis, B, Rooney, CM, Zhang, L, Mahendravada, A, Foster, AE, et al. (2009). T 376 lymphocytes coexpressing CCR4 and a chimeric antigen receptor targeting CD30 have 377 improved homing and antitumor activity in a Hodgkin tumor model. Blood 113: 6392-6402. 378 11. Li, N, Tang, N, Cheng, C, Hu, T, Wei, X, Han, W, et al. (2020). Improving the anti-solid tumor 379 efficacy of CAR-T cells by inhibiting adenosine signaling pathway. Oncoimmunology 9: 380 1824643. 381 12. Tang, N, Cheng, C, Zhang, X, Qiao, M, Li, N, Mu, W, et al. (2020). TGF-beta inhibition via 382 CRISPR promotes the long-term efficacy of CAR T cells against solid tumors. JCI Insight 5. 383 13. Markley, JC, and Sadelain, M (2010). IL-7 and IL-21 are superior to IL-2 and IL-15 in promoting 384 human T cell-mediated rejection of systemic lymphoma in immunodeficient mice. Blood 115: 385 3508-3519. 386 14. Alizadeh, D, Wong, RA, Yang, X, Wang, D, Pecoraro, JR, Kuo, CF, et al. (2019). IL15 Enhances 387 CAR-T Cell Antitumor Activity by Reducing mTORC1 Activity and Preserving Their Stem Cell 388 Memory Phenotype. Cancer Immunol Res 7: 759-772. 389 15. Krenciute, G, Prinzing, BL, Yi, Z, Wu, MF, Liu, H, Dotti, G, et al. (2017). Transgenic Expression 390 of IL15 Improves Antiglioma Activity of IL13Ralpha2-CAR T Cells but Results in Antigen Loss 391 Variants. Cancer Immunol Res 5: 571-581. 392 16. Waldman, AD, Fritz, JM, and Lenardo, MJ (2020). A guide to cancer immunotherapy: from T 393 cell basic science to clinical practice. Nat Rev Immunol 20: 651-668.

15 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

394 17. Heczey, A, Louis, CU, Savoldo, B, Dakhova, O, Durett, A, Grilley, B, et al. (2017). CAR T Cells 395 Administered in Combination with Lymphodepletion and PD-1 Inhibition to Patients with 396 Neuroblastoma. Mol Ther 25: 2214-2224. 397 18. Rafiq, S, Yeku, OO, Jackson, HJ, Purdon, TJ, van Leeuwen, DG, Drakes, DJ, et al. (2018). 398 Targeted delivery of a PD-1-blocking scFv by CAR-T cells enhances anti-tumor efficacy in vivo. 399 Nat Biotechnol 36: 847-856. 400 19. Hu, B, Zou, Y, Zhang, L, Tang, J, Niedermann, G, Firat, E, et al. (2019). Nucleofection with 401 Plasmid DNA for CRISPR/Cas9-Mediated Inactivation of Programmed Cell Death Protein 1 in 402 CD133-Specific CAR T Cells. Hum Gene Ther 30: 446-458. 403 20. Rupp, LJ, Schumann, K, Roybal, KT, Gate, RE, Ye, CJ, Lim, WA, et al. (2017). CRISPR/Cas9- 404 mediated PD-1 disruption enhances anti-tumor efficacy of human chimeric antigen receptor 405 T cells. Sci Rep 7: 737. 406 21. Ren, J, Liu, X, Fang, C, Jiang, S, June, CH, and Zhao, Y (2017). Multiplex Genome Editing to 407 Generate Universal CAR T Cells Resistant to PD1 Inhibition. Clin Cancer Res 23: 2255-2266. 408 22. Odorizzi, PM, Pauken, KE, Paley, MA, Sharpe, A, and Wherry, EJ (2015). Genetic absence of 409 PD-1 promotes accumulation of terminally differentiated exhausted CD8+ T cells. J Exp Med 410 212: 1125-1137. 411 23. Corral, LG, Haslett, PA, Muller, G, Chen, R, Wong, L, Ocampo, CJ, et al. (1999). Differential 412 Cytokine Modulation and T Cell Activation by Two Distinct Classes of Thalidomide Analogues 413 That Are Potent Inhibitors of TNF-a. J Immunol 163: 380-386. 414 24. Gandhi, AK, Shi, T, Li, M, Jungnelius, U, Romano, A, Tabernero, J, et al. (2013). 415 Immunomodulatory effects in a phase II study of lenalidomide combined with cetuximab in 416 refractory KRAS-mutant metastatic colorectal cancer patients. PLoS One 8: e80437. 417 25. Wu, L, Adams, M, Carter, T, Chen, R, Muller, G, Stirling, D, et al. (2008). Lenalidomide 418 enhances natural killer cell and monocyte-mediated antibody-dependent cellular cytotoxicity 419 of rituximab-treated CD20+ tumor cells. Clin Cancer Res 14: 4650-4657. 420 26. Schafer, PH, Gandhi, AK, Loveland, MA, Chen, RS, Man, HW, Schnetkamp, PP, et al. (2003). 421 Enhancement of cytokine production and AP-1 transcriptional activity in T cells by 422 thalidomide-related immunomodulatory drugs. J Pharmacol Exp Ther 305: 1222-1232. 423 27. Dredge, K, Marriott, JB, Todryk, SM, Muller, GW, Chen, R, Stirling, DI, et al. (2002). Protective 424 antitumor immunity induced by a costimulatory thalidomide analog in conjunction with 425 whole tumor cell vaccination is mediated by increased Th1-type immunity. J Immunol 168: 426 4914-4919. 427 28. Galustian, C, Meyer, B, Labarthe, MC, Dredge, K, Klaschka, D, Henry, J, et al. (2009). The anti- 428 cancer agents lenalidomide and pomalidomide inhibit the proliferation and function of T 429 regulatory cells. Cancer Immunol Immunother 58: 1033-1045. 430 29. Gandhi, AK, Kang, J, Havens, CG, Conklin, T, Ning, Y, Wu, L, et al. (2014). Immunomodulatory 431 agents lenalidomide and pomalidomide co-stimulate T cells by inducing degradation of T cell 432 repressors Ikaros and Aiolos via modulation of the E3 ubiquitin ligase complex CRL4(CRBN.). 433 Br J Haematol 164: 811-821. 434 30. Kroenke, J, Udeshi, ND, Narla, A, Grauman, P, Hurst, SN, McConkey, M, et al. (2014). 435 Lenalidomide Causes Selective Degradation of IKZF1 and IKZF3 in Multiple Myeloma Cells. 436 Science 343: 301-305.

16 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

437 31. Wang, Z, Zhou, G, Risu, N, Fu, J, Zou, Y, Tang, J, et al. (2020). Lenalidomide Enhances CAR-T 438 Cell Activity Against Solid Tumor Cells. Cell Transplantation 29. 439 32. John, LB, and Ward, AC (2011). The Ikaros gene family: transcriptional regulators of 440 hematopoiesis and immunity. Mol Immunol 48: 1272-1278. 441 33. Kathrein, KL, Lorenz, R, Innes, AM, Griffiths, E, and Winandy, S (2005). Ikaros induces 442 quiescence and T-cell differentiation in a leukemia cell line. Mol Cell Biol 25: 1645-1654. 443 34. Ng, SY, Yoshida, T, and Georgopoulos, K (2007). Ikaros and chromatin regulation in early 444 hematopoiesis. Curr Opin Immunol 19: 116-122. 445 35. Jager, R, Gisslinger, H, Passamonti, F, Rumi, E, Berg, T, Gisslinger, B, et al. (2010). Deletions of 446 the transcription factor Ikaros in myeloproliferative neoplasms. Leukemia 24: 1290-1298. 447 36. Zhong, XS, Matsushita, M, Plotkin, J, Riviere, I, and Sadelain, M (2010). Chimeric antigen 448 receptors combining 4-1BB and CD28 signaling domains augment PI3kinase/AKT/Bcl-XL 449 activation and CD8+ T cell-mediated tumor eradication. Mol Ther 18: 413-420. 450 37. Sadelain, M, Brentjens, R, and Riviere, I (2013). The basic principles of chimeric antigen 451 receptor design. Cancer Discov 3: 388-398. 452 38. Su, S, Hu, B, Shao, J, Shen, B, Du, J, Du, Y, et al. (2016). CRISPR-Cas9 mediated efficient PD-1 453 disruption on human primary T cells from cancer patients. Sci Rep 6: 20070. 454 39. Seo, W, Jerin, C, and Nishikawa, H (2021). Transcriptional regulatory network for the 455 establishment of CD8 + T cell exhaustion. Exp Mol Med 53: 202-209. 456 40. Pearce, EL, Mullen, AC, Martins, GA, Krawczyk, CM, Hutchins, AS, Zediak, VP, et al. (2003). 457 Control of effector CD8+ T cell function by the transcription factor Eomesodermin. Science 458 302: 1041-1043. 459 41. Li, G, Yang, Q, Zhu, Y, Wang, HR, Chen, X, Zhang, X, et al. (2013). T-Bet and Eomes Regulate 460 the Balance between the Effector/Central Memory T Cells versus Memory Stem Like T Cells. 461 PLoS One 8: e67401. 462 42. Wilson, RJ, Thomas, CD, Fox, R, Roy, DB, and Kunin, WE (2004). Spatial patterns in species 463 distributions reveal biodiversity change. Nature 432: 393-396. 464 43. Bao, S, Wu, Q, McLendon, RE, Hao, Y, Shi, Q, Hjelmeland, AB, et al. (2006). Glioma stem cells 465 promote radioresistance by preferential activation of the DNA damage response. Nature 466 444: 756-760. 467 44. Liu, G, Yuan, X, Zeng, Z, Tunici, P, Ng, H, Abdulkadir, IR, et al. (2006). Analysis of gene 468 expression and chemoresistance of CD133+ cancer stem cells in glioblastoma. Mol Cancer 5: 469 67. 470 45. Zhu, X, Prasad, S, Gaedicke, S, Hettich, M, Firat, E, and Niedermann, G (2014). Patient-derived 471 glioblastoma stem cells are killed by CD133-specific T cells but stimulate T cell aging markers 472 CD57. oncotarget 6: 171-184. 473 46. Patro, R, Duggal, G, Love, MI, Irizarry, RA, and Kingsford, C (2017). Salmon provides fast and 474 bias-aware quantification of transcript expression. Nat Methods 14: 417-419. 475

476

477

17 bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

478

Figure 1. Basic features of IKZF3 KO CAR T cells targeting HER2 (A) Successfully edited IKZF3 alleles in the CAR T populations. For CAR T cells derived from each donor, the targeted region was PCR-amplified and TA-cloned before 19 bacterial colonies were picked for Sanger sequencing. The values in the brackets indicate the numbers of deleted/inserted base pairs and the frequencies of the occurrence of the alleles among the 19 samples. PAM sites for the dual gRNAs are highlighted. (B) Western blot showing reduced IKZF3 expression in the crude edited population. (C) Relative CAR T cell numbers at the end of the expansion. Values are mean +/- SD (n= 3 technical repeats, *p<0.05, **p<0.01, ***p<0.001). (D-G) FACS analysis showing little effect of IKZF3 KO on CAR expression (based on anti-Myc staining; D-E) and CAR T cell subset composition (based on CD45RO and CD62L expression; F-G). (H-I) Increased CD69 expression frequency in KO CAR T cells. bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 2. IKZF3 KO enhances HER2-CAR T activity in vitro (A) Cytolysis of the target cells (luciferase-expressing MDA-MB-453) after 48 h co-incubation with CAR T cells at indicated effector to target (E/T) ratios. (B) Relative cytokine concentration in the culture media after 24 h co-incubation. (C) Proliferation of CFSE-labeled CAR T cells after 7-day culture in the presence (red) or absence (black) of lethally irradiated MDA-MB-453 cells. The values inside the plots are the “Percent Divided” (top) and the “Division Index” (bottom; see Materials and Methods). (D) Real time monitoring of cytolysis using impedance-based assay for 180 h. MDA-MB-453 cells, attached to the bottom of electrode plate, were cultured with CAR T cells at 1:1 (top) or 1:2 (bottom) effector to target ratio, and cell index (reflecting changes in impedance caused by the loss of tumor cells) measured every 15 min. (E) Cytokine concentration in the media at the end of 180 h incubation. Values are mean+/- S.D (n=3 technical repeats). Groups were compared through two-way ANOVA or two-tailed unpaired t-test. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. IL-2 was undetectable (not shown). bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 3. IKZF3 KO enhances HER2-CAR T activity in vivo Firefly luciferase-expressing MAD-MB-453 cells (1.5 x 106) were inoculated subcutaneously in the dorsal flank of right foreleg of the NPSG mice (5 mice/group) before intravenous injection of CAR T cells (5 x 106) 6 days later. Bioluminescence was imaged at various times thereafter (A), the luminescence intensity was quantified (B), and mice survival recorded (C). Groups were compared through two-way ANOVA. Error bars, s.d.; *p<0.05. bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 4. Transcriptome changes caused by IKZF3 KO (A) Differentially expressed genes (DEGs) revealed by RNA-seq, shown in heatmap (left) and MA- plot (right). DEGs are defined as the genes whose expression are changed by at least 1.5x (q<=0.1). RNA-seq was performed twice on different dates, using the CAR T cells derived from Donor 0710 (biological replicates). (B) Enriched pathways in the upregulated (top) and downregulated (bottom) DEGs, as analyzed by DAVID. GO, Gene Ontology; KEGG, Kyoto Encyclopedia of the Genes and Genome; FDR, False Discovery Rate from Benjamini and Hochberg. (C) Expression levels (in TPM) of selected genes from RNA-seq data. The values are mean+/- SD (n=2 biological replicates). bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 5. Basic features of IKZF3-deficient CD133-specific CAR T cells (A-I) Same as Fig. 1A-I, except that the CAR T cells targeting CD133 were analyzed bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Figure 6. 133K3-CAR T cells showed improved antitumor efficacy compared with 133-CAR T cells in vitro (A-E) Same as Fig. 2A-E except that the tumor cells were U251 cells expressing firefly luciferase and CD133, and T cells expressed CD133-specific CAR. bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Fig. S1. IKZF3 KO enhances HER2-CAR T activity in vitro Shown is a biological replicate for the experiment described in Fig. 2A-D, except that in Fig. S1B, only one E/T ratio was tested. bioRxiv preprint doi: https://doi.org/10.1101/2021.06.18.449074; this version posted June 19, 2021. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Fig. S2. F133K3-CAR T cells showed improved antitumor efficacy compared with 133- CAR T cells in vitro Shown is a biological replicate for the experiment described in Fig. 6A-D except that in Fig. S2B, only a single E/T ratio was tested.